deegree community space 2012 planning
What is the (intention of) deegree community space?
deegree community space is an event dedicated to the community of the deegree initiative. Communication is the most important aspect here, while information is the second. The new event name deegree community space shall support this intention - both by explicitly naming the community and by approaching the open space concept. The project steering committee will host this event supported by the technical management committee. We see this as a major opportunity to make deegree a better community initiative.
In contrast to events like the past deegree days here we need to have and to arrange two-way communication in the first place. And looking at community building it is obviously most important that community members get to know each other. That's why we think that important aspects of the event are:
- Fixed time frame (10 - 18 h) with a schedule grid (will be prepared by the PSC)
- Fill the grid at the beginning of the event, but not entirely
- All in one room, most of the time
- Review the schedule (half-way or somewhere in between)
- Have room and time for breakouts in small groups, if needed
Potential contributions can be presented on the deegree-users mailing list. They can be discussed there. The following list provides an overview of the contributions proposals we have so far:
- Markus Schneider: The deegree workspace: Open issues (Presentation/discussion)
- Volker Grabsch: Packaging deegree for Debian (Presentation)
- N.N.: deegree doctor - the cure for your deegree problems (Desk)
- N.N.: Mini code sprint (no focus yet) (Breakout)
- PSC: Vision of the deegree initiative (Discussion)
- All (?): deegree show cases, a standup trade fair of deegree-based applications (some Desks)
- TMC: deegree web services handbook (Discussion)
- TMC: deegree project infrastructure (Discussion)
- Herman Assink: Performance and scalibility of deegree (Presentation/Discussion)
- Reijer Copier: Using deegree in a Java enterprise environment (Presentation/Discussion)
- Just van den Broecke: Preparing your rich GML data for deegree: the ETL Challenge (Presentation/Discussion)
- TMC: deegree stability guidelines for configuration and API (Discussion)
- TMC: deegree roadmap/actions for the 3.2 release (Discussion)
You may find more information about these contribution proposals below the line.
Currently available information on contribution proposals
0.1. MarkusSchneider: The deegree workspace: Open issues (Presentation/discussion)
Configuration of deegree 3 is based on a unified, extensible and resource-oriented approach. Regardless if you want to configure map layers, feature data sources or metadata access: Every configuration aspect is a defined by the deegree workspace and the same overall principles are applied everywhere. Physically, the deegree workspace is a directory of configuration files with a well-defined layout. It evolved from the experience that was gained with earlier versions of deegree (deegree 1 and deegree 2). The following subjects will be covered:
- What is the deegree workspace?
- General workspace directory structure
- Types of workspace resources
- Open issue: JDBC resources (connection parameters or pools)
- Open issue: Integration of metadata
0.2. VolkerGrabsch: Packaging deegree for Debian
At m-click we have a demand for clean deegree packages that integrate well into our Debian systems.
This presentation explains how we are creating those Debian packages, and how we distribute them through our custom APT repository. The advantages and limitations of our approach are discussed, as well as possible future developments.
Our repository is publicly available for demonstration, and we'd love to enable the deegree project to setup and maintain their own official repository.
0.3. JustVanDenBroecke: Preparing your rich GML data for deegree: the ETL Challenge
Mapping and serving rich GML schema's like for INSPIRE is now possible in various ways with deegree3 (see Markus' tutorial on the SQL Feature Store above). But how do we prepare our source data such that it can be stored for and served by deegree? In particular for INSPIRE we have to start with our original, national source data that often has a completely different model, format and/or projection. The problem of data transformation is off course recurring in the geospatial domain. Most of us have to deal with 'Spatial ETL', where ETL stands for Extract, Transform, Load. There are various tools both closed and open source. Of the latter category GDAL/OGR with tools like ogr2ogr are well known. But dealing with rich/complex GML creates extra ETL-challenges. Examples are INSPIRE datamodels but also many national datasets like Top10NL (Topography) and BAG (Addresses) in the Netherlands and ALKIS in Germany . The problem is at least twofold: dealing with extremely large quantities (millions of objects) of GML data either as input and/or output and dealing with complex model transformations. And finally we have to devise a strategy for combining our ETL with a deegree database storage mode. Apart from deegree serving we also need to generate downloadable GML files for INSPIRE Atom feeds. Existing Open Source ETL tools like GeoKettle en Talend Geospatial were investigated but we wanted something more lightweight, "batch-like", and integrate closely with deegree.
In 2009-2011 the Dutch Kadaster successfully prototyped INSPIRE data harmonization and network services using Free and Open Source Software (FOSS), including deegree for WFS. This work was carried out initially within European projects like ESDIN (EuroGeographics). All results from the Dutch Kadaster like documentation and software are published through the ongoing FOSS-INSPIRE project http://inspire-foss.org. In this project we combined several ETL tools like GDAL (ogr2ogr) and XSLT (model transformation) but in a rather ad-hoc fashion.
Recently, as an offspring from the INSPIRE-FOSS project a generic ETL-framework called SETL https://github.com/justb4/setl is now developed. The "S" stands for Simple, Streaming or Spatial. SETL specificly targets complex transformations like required for INSPIRE and complex ("rich") GML Application Schema's in general. Based on Unix-principles like Pipes and Filters and Java Spring Declarative Programming, users can construct "ETL-Chains" with SETL without programming. Integration with deegree is effected through SETL-modules like Deegree Blobstore Output and WFST-Output. SETL, written in Python , builds on and integrates proven FOSS tools and libraries like GDAL/OGR and XSLT (via libxslt2) natively. By using techniques like I/O-streaming and "GML-Splitting", both databases and GML files with millions of features have been transformed successfully. The combination SETL+deegree will be proposed as a complete INSPIRE solution.